Content uploaded by Colin Campbell
Author content
All content in this area was uploaded by Colin Campbell on Feb 19, 2020
Content may be subject to copyright.
From data to action: How marketers can
leverage AI
Colin Campbell
a,b,
*, Sean Sands
b
, Carla Ferraro
b
,
Hsiu-Yuan (Jody) Tsao
c
, Alexis Mavrommatis
b,d
a
School of Business, University of San Diego, 5998 Alcala Park, San Diego, CA 92101,
U.S.A.
b
Swinburne University of Technology, Hawthorn, VIC 3122, Australia
c
National Chung Hsing University, Taiwan
d
EADA Business School, Barcelona, Spain
KEYWORDS
Artificial intelligence;
Machine learning;
Marketing function;
Marketing mix;
Consumer engagement;
Customer experience;
Customer journey
Abstract Artificial intelligence (AI) is at the forefront of a revolution in business
and society. AI affords companies a host of ways to better understand, predict,
and engage customers. Within marketing, AI’s adoption is increasing year-on-year
and in varied contexts, from providing service assistance during customer interac-
tions to assisting in the identification of optimal promotions. But just as questions
about AI remain with regard to job automation, ethics, and corporate responsibil-
ity, the marketing domain faces its own concerns about AI. With this article, we
seek to consolidate the growing body of knowledge about AI in marketing. We
explain how AI can enhance the marketing function across nine stages of the mar-
keting planning process. We also provide examples of current applications of AI in
marketing.
ª2019 Kelley School of Business, Indiana University. Published by Elsevier Inc. All
rights reserved.
1. Artificial intelligence: Seeing the
forest for the trees
New technologies have the potential to disrupt
consumer behavior, management processes, and
organizational strategy (Evans, 2017). Artificial
intelligence (AI) is one such disruptive technology,
affecting a diverse range of industries from health
care to retail. AI involves the development of
valuable, automated solutions to problems that
would require the intervention of intelligence if
completed by humans (Martı
´nez-Lo
´pez & Casillas,
2013; Negnevitsky, 2004). AI is increasingly un-
derpinning a vast array of customer-brand in-
teractions. To optimize the customer experience,
* Corresponding author
E-mail addresses: colincampbell@sandiego.edu (C. Campbell),
ssands@swin.edu.au (S. Sands), cferraro@swin.edu.au (C.
Ferraro), jodytsao@dragon.nchu.edu.tw (H.-Y.(J.) Tsao),
amavrommatis@eada.edu (A. Mavrommatis)
https://doi.org/10.1016/j.bushor.2019.12.002
0007-6813/ª2019 Kelley School of Business, Indiana University. Published by Elsevier Inc. All rights reserved.
Business Horizons (xxxx) xxx, xxx
Available online at www.sciencedirect.com
ScienceDirect
www.journals.elsevier.com/business-horizons
for instance, many firms employ AI and machine
learning (ML) to predict customer demands, assist
frontline service employees in serving customers,
and allow simple service queries to be answered by
bots.
AI is creating efficiencies on an unprecedented
scale, leading to automated and interconnected
business processes that have diverse implications
for a wide range of business functions, including
marketing. As a result, marketing managers
need to consider reshaping internal capabilities,
rethinking strategies, and how customer in-
teractions might be transformed. While some firms
are striving to lead all business decisions with
consideration of AI (Baker, 2017; Wolska, 2017),
others are struggling to see the forest for the trees
and to navigate AI adoption. The purpose of this
article is to consolidate the growing body of
knowledge about AI in the field of marketing. Such
understanding is critical given the frequent digital
interactions between brands and customers
(Bughin, McCarthy, & Chui, 2017), and will assist
marketing managers in considering the adoption of
AI.
AI offers a range of opportunities for the field
of marketing (Martı
´nez-Lo
´pez & Casillas, 2013).
Indeed, research in the U.S. suggests a large
proportion of business-to-consumer marketers are
already taking advantage of AI and ML (Narrative
Science, 2018). For those organizations that
employ AI, it is primarily being used to target
audiences, make product recommendations, and
optimize advertising campaigns (Blueshift, 2018).
But significant potential remains for marketers to
leverage advanced AI capabilities, with only a
small proportiondjust 6%dreporting the use of
such capabilities, which include personalizing
campaigns, collaborative filtering, and predictive
models (Blueshift, 2018). AI should be a consid-
eration for all marketing managers as it repre-
sents the highest growth of any technology in
marketing (Salesforce, 2017), is expected to in-
crease in use (Columbus, 2018), and is predicted
to have a $40 billion effect on marketing by 2025
(Reavie, 2018).
Although the vast majority of marketing man-
agers believe AI has a revolutionizing potential,
many are still unaware of the magnitude of the
benefits or unsure how they can adopt AI to
improve marketing (Demandbase, 2016; Reavie,
2018). In addressing these questions, we first pro-
vide a brief history of AI before offering a more in-
depth analysis of AI and ML. Next, we outline the
opportunities for applying AI to marketing strat-
egy, including examples of current applications.
Finally, we conclude with guidelines for what it
takes to succeed in an AI-first business environ-
ment and provide thoughts for the potential
growth of AI within the marketing discipline.
2. A primer: What marketers need to
know about AI
2.1. AI is more than an evolution in statistics
The amount of data generated today by both
humans and machines far outpaces humans’ ability
to absorb, interpret, and make complex decisions
on the basis of that data (Hurwitz, Kaufman, &
Bowles, 2015). AI can help address this problem.
The rapid development of AI, coupled with cloud-
based resources, connectivity, and platform-
based business models (those focused on bringing
parties together on a platform to exchange prod-
ucts and services for money), is leading to auto-
mated and interconnected business processes that
have implications for customers and other stake-
holders. For marketers, AI affords strong oppor-
tunities for innovative humanemachine
integration (Rust & Huang, 2014), with applica-
tions in advertising, strategy, logistics, and
customer experience, to mention a few. For
instance, AI can provide valuable insights about
finding the right consumers, engaging with cus-
tomers, and conducting return-on-investment
analysis.
Because of the vast availability of data and the
advent of increasingly cheaper and faster
computing power, AI and ML also afford insights
beyond those of traditional statistical methods. AI
and ML do not apply rigid assumptions about the
problem, nor about data distributions in general;
they employ many approaches and techniques to
find a solution. While AI and ML techniques can be
based on deductive or inductive learning, the
benefit of inductive learning is that not much prior
knowledge is needed about the problem or the
data. In contrast, traditional statistics is based on
deductive learning (which relies on prior knowl-
edge about data) and thus makes tight assump-
tions about the problem and nature of the data
(Teboul, 2018). In other words, AI and ML methods
enable learning from data and discard assumptions
attached to statistical methodologies.
Given the business benefits of AI, there are
myriad customer-focused applications increasingly
observable in a range of industries, including
retail, finance, healthcare, education, trans-
portation, and communications. For example, vir-
tual bots are turning customer service into self-
service (Fluss, 2017), big-data AI applications are
2 C. Campbell et al.
replacing portfolio managers (Javelosa, 2017), and
social robots are replacing human greeters to
welcome customers in service sectors (Choudhury,
2016). Indeed, growth in AI development and
deployment is not expected to slow. McKinsey
forecasts that by 2020, U.S. customers will manage
85% of their brand relationships without human
interaction (Baumgartner, Hatami, & Valdivieso,
2016).
2.2. AI has different building blocks
AI is typically defined as technology that enables
machines to learn from experience and perform
human-like functions (Marr, 2018). We focus on ML
because it underpins the functionalities AI affords
(McCorduck, 2009). In broad terms, ML refers to
software that is able to learn how to accomplish a
task without explicit instruction. ML algorithms
detect patterns and learn how to make predictions
and recommendations by processing data and ex-
periences rather than by receiving explicit pro-
gramming instruction. ML is a powerful tool for
mining large sets of data, providing marketers the
opportunity to gain new insights into consumer
behavior and to improve the performance of
marketing operations (Cui, Wong, & Lui, 2006). ML
is used in a wide variety of applications that power
many aspects of modern society, including web
searches, content filtering on social networks, and
e-commerce recommendation systems (LeCun,
Bengio, & Hinton, 2015). ML has emerged as the
method of choice for developing practical soft-
ware for computer vision, speech recognition,
natural language processing, robot control, and
other applications (Jordan & Mitchell, 2015).
Table 1 provides a brief overview of AI and its sub-
domains of ML and deep learning, along with some
key terms.
Machine learning can be subdivided into
different forms, with the three key paradigms
shown in Table 2. A discussion of each form follows
(Chiu et al., 2018; Das, Doppa, Kim, Pande, &
Chakrabarty, 2015; Davenport & Kirby, 2016).
2.2.1. Supervised ML
The most widely used ML methods are supervised
learning methods (Jordan & Mitchell, 2015). An
example of supervised learning would involve a
system learning the difference between a koala
Table 1. AI subdomains and key terminology
Term Definition
AI and AI Subdomains
Artificial intelligence (AI) The broad class of technology developed with the objective of collecting data in
order to solve problems or make decisions.
Machine learning (ML) An application of AI that provides systems the ability to automatically learn and
improve from experience without being explicitly programmed.
Deep learning A form of neural network that develops understanding by building successively
more abstract representations of a data set. This occurs by separating a data set
into different layers of abstraction or transformation and then searching for
patterns, first within each layer and then between them.
Key AI Terminology
Neural network A form of AI model designed to approximate how the human brain operates. It
works by breaking problems down into smaller components and then iteratively
solving them, building on the findings of earlier stages.
Target leakage When an AI model accidentally includes information that would not be known at
the time of prediction. Because such information is often highly predictive of the
outcome that is trying to be predicted, this produces overly optimistic estimates
during model training of out-of-sample performance.
Feature engineering Cleaning and manipulation of data that are input into an AI model. This includes
simple tasks such as ensuring data are coded correctly (e.g., attaching days of the
week to dates), as well more complex manipulations, such as creating
transformations of or interactions between variables.
Ensemble model A form of ML that combines several different models in order to improve
prediction. This can be accomplished by using blenders, when several different
models are run concurrently and their results are merged into a final prediction, or
by stacking, when models are employed sequentially with one model’s outputs
forming inputs into another model.
How marketers can leverage AI 3
and a kangaroo. By examining dozens of labeled
examples, the system can induce which features
are useful for distinguishing between the two an-
imals and thus improve its prediction. This same
approach is used to train chatbots to identify
common customer queries, and to train spam fil-
ters to identify unwanted emails.
More technically, supervised learning refers to
situations when ML algorithms see data that in-
cludes both a series of predictor variables as well
as an outcome. This data set can be split into
training and holdout parts. The ML algorithm can
then analyze the training set, looking for patterns
between predictor variables and the particular
outcome. Models relying on many different algo-
rithmic approaches can then be compared by
validating them using the holdout data. Supervised
ML is focused on trying to predict a particular
outcome and hinges on the existence of a data set
composed of examples of predictor variables and
known outcomes.
2.2.2. Unsupervised ML
While supervised learning involves analyzing data
and attempting to predict a particular outcome,
unsupervised learning involves analyzing data
without trying to predict anything. Unsupervised
learning is instead focused on understanding the
underlying structural properties of a data set in
order to discover useful representations of the
input without the need for labeled training data
(Jordan & Mitchell, 2015). Clustering, an unsu-
pervised learning approach, focuses on finding as-
sociations between observed data in the absence
of any explicit signs of association. Clustering en-
ables rules to be developed to classify future data.
Unsupervised ML can be used for segmenting cus-
tomers and markets, classification, and detecting
outliers. In sum, while unsupervised ML relies on a
data set to operate, it is distinguished by its focus
on finding structure rather than predicting a
particular outcome.
2.2.3. Reinforcement ML
Reinforcement ML refers to situations where an
existing data set does not exist, and an algorithm
learns by taking different actions and evaluating
their success or failure. In this instance, a learning
system doesn’t have a historical data set to draw
upon, so immediate and continuous feedback en-
ables the system to learn while building a data set.
An example of reinforcement ML is advertising on
Facebook with a conversion-tracking pixel
installed. When an ad is first developed and fligh-
ted, Facebook’s algorithm tests the ad across the
full spectrum of targeting. As sales success occurs,
the algorithm can analyze the data to better refine
its targets, possibly by concentrating on a certain
audience, during certain times of day, in certain
geographic locations, and using certain on-screen
placements. Reinforcement ML can also be used in
developing recommendation systems and opti-
mizing logistic flows.
2.2.4. Hybrid ML systems
Although the three ML paradigms help in under-
standing how ML operates, most current research
involves a blend of these forms (Jordan & Mitchell,
2015). This approach often blends or stacks what
are called ensemble models (see Table 1) in order
to improve their prediction. For instance, a model
might employ unsupervised ML to classify and
organize data, then relate these classifications
against an outcome using supervised ML. Likewise,
supervised ML can be used to identify useful pre-
dictors, which can then be refined with rein-
forcement learning to yield better predictions
going forward.
3. AI-enabled marketing: Opportunities
and applications
Advances in the field of big data provide marketers
the ability to collect and aggregate vast amounts
Table 2. Comparison of key ML paradigms
Supervised
learning
Unsupervised
learning
Reinforcement
learning
Task defined? Yes No Yes
Existing data needed? Yes Yes No
ML actively makes decisions and assesses
outcomes?
No No Yes
Occurs over time? No No Yes
4 C. Campbell et al.
of information, with the ultimate aim of turning
data into insight or actionable strategy. AI can
greatly assist marketers in this process by drawing
conclusions from unstructured data about causes
and effects within extremely large data sets. With
the ability to detect and extrapolate upon pat-
terns, AI can help marketers identify opportunities
and act upon them in real time. As a means to
provide an organizing framework and assist mar-
keting professionals in understanding and deter-
mining effective uses for AI, we structure our
discussion around nine components inherent to
strategic marketing (Wood, 2011). We outline how
AI might be leveraged at each stage, as well as the
data and techniques that might best serve each
stage. While we provide examples of current ap-
plications, it is important to note that the exam-
ples may have effects beyond a single stage
because AI applications are often broad in nature.
For instance, the use of AI to extract sentiment
from customers in social channels could offer
insight in a variety of stages. Table 3 provides a
summary of AI’s potential across each of the nine
stages of the marketing plan.
3.1. Analyzing the current situation
Stage 1 involves analyzing the current situation
and understanding macroenvironmental factors
that can affect the organization, its marketing,
and its stakeholders. At this step, marketing
managers strive to develop an understanding of
the current and future environment in which the
company operates (Chaffey, 2004; Chaffey &
Smith, 2012); managers can evaluate markets,
the opportunities within them, and the threats
arising from AI adoption decisions (Paschen, Pitt, &
Kietzmann, 2020). Several tools available to mar-
keters, such as SWOT or PESTLE analysis, can assist
them in understanding the specific markets they
operate in and the consumers they target. AI
techniques, including social listening, can glean
information on markets and consumers, particu-
larly in terms of satisfaction, purchasing patterns,
and product demand. From this perspective, AI
affords marketers the opportunity to identify
changes in competitor behavior (including pricing),
estimate product demand, and assess customer
sentiment (including customer satisfaction).
In analyzing the market, it is important for
marketing managers to focus on competing brands,
product alternatives, and available channels
within the entire category landscape. Categories
can develop rapidly with advancements in tech-
nology and available choices, leading to market
volatility. Social media and online forums afford
consumers the ability to research products most
suitable to their specific needs. Oculus360 (2018),
an AI-market and consumer-research agency, used
ML to investigate consumer discussions in online
forums. The research shows the importance of
analyzing online conversations concerning the
entire category rather than strictly focusing on
one’s own specific brand. Such insights are perti-
nent for broad situation analysis, providing a gauge
of how well consumer segments are being served
by one’s own brand as well as competitors’.
3.2. Understanding markets and customers
Stage 2 involves understanding markets and cus-
tomers, as well as gathering knowledge of micro-
environmental factors that specifically affect the
firm, including market-share trends, product/
category demand, and customer characteristics,
including needs, wants, behaviors, attitudes,
brand loyalties, and purchasing patterns. At this
stage, marketers aim to develop an understanding
of the specific markets they operate in and the
consumers they target, monitoring their behavior
to track the success of previous stages in terms of
key metrics. During this process, web analytics and
traditional market research (customer satisfac-
tion) are often engaged, with AI providing a vast
array of opportunities beyond these. For instance,
voice-of-customer programs are allowing mar-
keters to move beyond interview-based data to
also incorporate large amounts of unstructured
customer data. Medallia is one customer-
experience software provider that has integrated
AI capabilities to mine customer preferences and
data from the web, social media, mobile activity,
and contact-center interactions (Dunwoodie,
2018). In this context, data can be analyzed and
feedback provided in real time, allowing decisions
and actions to be taken immediately.
Furthermore, AI is expanding the available
sources of data that firms have access to and is
extending traditional satisfaction metrics. An
example of this was demonstrated at the 2019
Consumer Electronics Show in Las Vegas, where
software developer Neurodata Labs and robotics
manufacturer Promobot unveiled multimodal
emotion detection for customer-experience man-
agement (Ponce de Leon, 2019). This is an AI sys-
tem that is able to analyze a combination of
human activities (e.g., facial expressions, body
gestures, voice, eye movement, and heart rate)
and determine a consumer’s emotional state. The
technology is being tested by a Russian bank,
How marketers can leverage AI 5
Table 3. Marketing functions and AI potential
What AI Can Offer Data Requirements Examples of Current AI Implementations
Stage 1: Analyzing the current situation: Involves understanding macroenvironmental factors that can affect the organization, its marketing, and its stakeholders
Analysis, simplification, provision,
and understanding of large unstruc-
tured data sets
Identification of anomalies (events)
in the market
Recommender systems to identify
likely future events (e.g., areas of
potential growth)
Sentiment analysis
External data, including census data,
demographics, consumer confidence,
macromarket trends, third-party
data (e.g., news stories, stock prices,
home sales), social-media discussions
Sociocultural trends through online chatter
Stock-market prediction
Improved macroeconomic forecasting drawing on a
wider array of indicators than present models
Stage 2: Understanding markets and customers: Entails gathering knowledge of microenvironmental factors that specifically affect the firm, including market-share
trends, product/category demand, and customer characteristics: needs, wants, behaviors, attitudes, brand loyalties, and purchasing patterns
Identification of changes in compet-
itor behavior (e.g., pricing)
Estimation of product demand
Assessment of customer sentiment
(e.g., customer satisfaction, social
media sentiment analysis)
Internal data, including sales (cur-
rent and historical, sales of own
products), customer data (satisfac-
tion, attitudes, demographics, etc.),
market research (e.g., ad/promotion
testing); External data, including
market share, scanner data, sales
(sales of competitors’ brands, sea-
sonality, weather, holidays), social-
media comments, competitors’
pricing and product availability
(e.g., sales, stock-outs)
Synthesis and understanding of customer comments,
feedback, and interests
Market research using analysis of facial expressions,
eye movements, and audio comments
Mapping the complexity of the customer journey and
associated understanding of the effect of individual
components (ads, touchpoint, and influencers) along
the way
Detection of changes in competitor behavior
(e.g., pricing, distribution)
Under Armour uses AI to perform consumer-sentiment
analysis and social listening to understand what
customers think of the brand and where the gaps in
the market are
Stage 3: Segmenting, targeting, and positioning: Involves developing an understanding of customer segments and assisting with targeting and positioning decisions
Classification and clustering of cus-
tomers into distinct segments
Internal data, including loyalty and
sales information, customer willing-
ness to purchase, and brand
Development of much smaller segments (moving to-
ward true one-to-one marketing)
6 C. Campbell et al.
Estimation of the probability of
response to promotions
Improved targeting of ads
Product and brand recommendations
perceptions; External data, including
demographics, census data, and
location
Clustering of consumers using vast data sets
Predictive modeling to optimize targeting decisions
Target received notoriety in 2012 for their ability to
segment and target products to expectant mothers
(Duhigg, 2012)
Stage 4: Planning direction, objectives, and marketing support: Entails developing longer-term goals and associated short-term objectives to support larger
strategies
Provision of digital customer service
(e.g., chatbots)
Estimation of the responsiveness of
consumers to price changes and
promotions
Combination of information from the
macro- and microenvironments to
better inform marketing objectives
Identification of those most likely to
purchase
Internal data, including historical
data on areas of marketing (e.g.,
amounts, types, and locations of
ads/sales support) and associated
outcomes (e.g., site traffic, leads,
sales); External data, including
census data, demographics, con-
sumer confidence, macro-market
trends, and third-party data
(e.g., news stories, stock prices,
home sales)
Estimation of sales and revenues (e.g., Aviso, Data-
Robot) as well as expected expenses of new and
existing products
Estimation of the reactiveness of sales to ad spends,
price changes, etc.
Travel-booking website Expedia used Amazon Sage-
Maker AI to train its ML models to identify and high-
light the most attractive hotel images in its data set;
Expedia was then able to predict images that
increased click-through and purchases
Stage 5: Developing product strategy: Involves creation of the suite of products sold by a firm
Identification of gaps in the market
for new product development
Creation of more customized and
boutique products
Awareness of what is in style or
trendy and thus worth producing and
selling
Assistance with designing and pro-
ducing products customized to indi-
vidual consumers
Historical data on customers, their
purchases, and associated outcomes
(e.g., satisfaction, returns) in order
to create recommendations; Data-
bases of consumer profiles from
which to estimate new customers’
sizes/profiles depending on inputs;
Information on trending products,
topics, and styles from social media,
press articles, etc.
New product development (e.g., Choosy fast fashion)
based on trends identified through analysis of social
tagging
Hyper-individualized product customization
(e.g., Zozo’s clothes that are customized according to
AI-enabled customer measurements)
Stage 6: Developing pricing strategy: Revolves around determining pricing strategies to maximize sales
Estimation of consumer price elas-
ticity at both individual and collec-
tive levels
Both historical and real-time sales,
search, and price data on firm and
competitor products
Retailers such as Amazon use algorithms that auto-
matically increase prices in response to spikes in
demand
(continued on next page)
How marketers can leverage AI 7
Table 3 (continued)
What AI Can Offer Data Requirements Examples of Current AI Implementations
Provision of dynamic pricing
(e.g., surge pricing) and price
discrimination.
Detection of anomalies (e.g., errors
in pricing, fraud, or nonprofitable
customers).
Wise Athena leverages ML to analyze both own and
competitor prices to help firms make optimal price
and promotional decisions
Stage 7: Developing channels and logistics strategy: Involves determining logistics, distribution, and product stocking decisions
Prediction and optimization of dis-
tribution, inventory, store displays,
and store layouts (both brick-and-
mortar and online)
Enabling voice and visual search
Data at the store level (historical and
real-time sales, real-time inventory,
in-store and web traffic data) and
location level (local competitors,
demographics of local catchment);
Data on individual customers
(historical sales, search history, any
other customer-level data useful for
making product recommendations);
Historical customer service queries,
responses, and satisfaction scores
AI-driven stock and inventory management
(e.g., Afesh)
Merchandising based on AI prediction (e.g., Celect)
Recommendation engines that show people what they
want to see (e.g., Reflektion)
Use of AI-driven camera analysis to speed payment
and optimize store layout and design. For instance,
IMAGR makes SmartCart, an ordinary grocery cart with
an AI computing video camera. The device tracks
what goes into the cart, tallies the total along the
way, and syncs that with payment information on the
shopper’s mobile phone
AI-enabled visual search (e.g., GrokStyle)
SapientRazorfish’s COSMOS platform gleans informa-
tion about a customer’s profile and purchasing history.
Sephora can then notify customers either by email,
direct mail, or SMS when their favorite products are in
stock or on sale, and even when customers are near
one of their retail stores
Stage 8: Developing marketing communication and influence strategy: Focuses on serving customers the right promotion at the right time
Creation of different ads depending
on permutations of content, and on
related words
Both historical and real-time data on
ads, including their content (both
text and images), placement, and
performance; Information on
Granify uses ML to identify shoppers ready to abandon
a cart and to make real-time offers to encourage
purchase
8 C. Campbell et al.
Development of individualized pro-
motional offers and ads
Running of AI-driven A/B testing
Optimization of ad placement
Reduction in cart abandonment
Contextual ad targeting
Optimization of ad retargeting
Keyword bidding and cost reduction
Automation and personalization of
content creation
potential ad placements (e.g., costs,
audience characteristics); Real-time
data on customer behavior at all
points along the consumer journey
Watson Ads Omni enables marketers to deploy AI-
powered interactive ad placements. Lego was the first
brand to use Watson Ads Omni on Black Friday, 2018
(Sweeney, 2018)
Chinese video start-up Viscovery is developing AI
technology that enables brands to display ads in
videos specific to the content being watched (Caygill,
2017)
Stage 9: Planning metrics and implementation control: Involves identifying performance metrics, monitoring them, and then taking any needed corrective actions
Better prediction of expected reve-
nues and profits, as well as their
variability
Identification of metrics linked to key
outcomes
Prediction of the effect of correct
actions and, in some cases, auto-
matically taking steps to diagnose,
correct, and improve on poor results
Both historical and real-time sales
and marketing performance data;
Real-time data facilitates diagnosing
problems, while historical data
enables prediction of corrective
actions
US financial software company Intuit uses ML enabled
by Amazon Web Services for real-time fraud analysis
Identification of promotional and pricing mistakes
How marketers can leverage AI 9
Rosbank, at their call center (Vilar, 2019). Data is
collected from customers, including the number of
pauses in speech, changes in voice volume, and
the total conversation time, and converted in real
time to a customer-satisfaction metric.
3.3. Segmenting, targeting, and positioning
Stage 3 involves segmenting, targeting, and posi-
tioning, which involves developing an under-
standing of customer segments and assisting
marketing managers in their targeting and posi-
tioning decisions. In this process, marketers seek
to group consumers according to certain criteria,
enabling precise targeting of messages and the
creation of brands and products that can best ap-
peal to each segment. AI not only helps to predict
customers’ intent but can help marketers segment
customers into more refined groups. Given the
large heterogeneity of consumer tastes and pref-
erences, the potential of segmentation is
immense, from tailoring promotions and ads to
making better product and brand recommenda-
tions. For example, in an attempt to improve its
customer targeting, Harley-Davidson teamed up
with U.S. marketing firm Adgorithms to use its
platform, Albert. Albert uses AI and ML to auto-
mate and simplify marketing planning. Harley-
Davidson provided Albert information on past
customers, enabling the machine to create lists of
similar audiences and match people who resemble
their current buyers. The platform was responsible
for 40% of Harley-Davidson’s motorcycle sales and
led to a ninefold increase in inbound calls (Power,
2017). AI is also being used to create personalized
content targeted to individual consumers. In 2018,
Wimbledon partnered with IBM to create a suite of
tools and services, including two chatbots, on-site
augmented-reality experiences, and a scoring and
insights app (SlamTracker) that is tailored to the
fan engaging with it (Moore, 2018).
Despite the benefits of using AI for segmenting,
targeting, and positioning, marketers should be
aware of the dangers of discrimination through AI.
While businesses inherently discriminate in terms
of to whom they offer products and services, AI
can lead to unintended and illegal price discrimi-
nation through its emphasis on targeting different
audiences. In the EU, it is now considered
discriminatory behavior when algorithms are used
to set prices on the basis of observable group
characteristics. While a group may exhibit a gen-
eral behavior, any individual within that group may
not, and that person would therefore be discrimi-
nated against if observable group trends were used
to determine pricing (Newell & Marabelli, 2015).
3.4. Planning direction, objectives, and
marketing support
Stage 4 involves planning direction, objectives,
and marketing support. This stage entails the
development of longer-term goals and associated
short-term objectives to support larger strategies.
Wood (2011) noted that growth strategies,
nongrowth strategies, objectives (marketing,
financial, and societal), and customer service are
important considerations within this stage. To
assist growth strategies, AI and chatbots can be
integrated into apps or social media to encourage
consumer purchasing. One example is the Star-
bucks Barista bot for Facebook Messenger. The bot
lets users place coffee orders via either voice
command or a messaging interface.
One of the most common applications of AI in
this stage is in customer service. In the United
States, just 24% of customer service teams are
using AI, but 56% report actively looking for ways
to use AI (Salesforce, 2019). Including such appli-
cations as chatbots and text and voice analysis, AI
use in customer service is forecasted to increase
by 143% between 2019 and 2021 (Salesforce,
2019). Today, chatbots are being employed in
customer service to address most simple queries.
Chatbots reduce customer service costs, but their
effect on customer satisfaction can be varied.
Many consumers still prefer to speak with human
agents for more complicated requests. In the U.K.,
nearly 50% prefer a human over a chatbot, and in
the U.S., 40% prefer a human over a chatbot
(Elliott, 2018). Despite consumer preferences for
human interactions, AI can still provide back-end
support in customer service. For instance, AI can
act to assign agents to customers. This process can
ensure that customers are connected with an
agent who has the expertise to address the cus-
tomer’s needs. One such use would be classifica-
tion systems that use natural-language processing
to identify the problems customers report. By
better matching agents with customers, AI can
streamline the interaction and preserve value for
firms.
3.5. Developing product strategy
Stage 5 involves the development of product
strategy (i.e., the creation of the suite of products
sold by a firm). At this stage, marketers use their
understanding of target consumers and of the
intended position for the brand to help develop
successful products. This typically involves de-
cisions about a product’s design, features, quality,
10 C. Campbell et al.
and customization. Opportunities for AI assistance
in product strategy include identifying gaps for
new product development, facilitating the pro-
duction of products customized to consumers’
specifications, and assisting with product delivery
and logistics. AI can also identify which products to
manufacture. Fast-fashion brand Choosy draws
fashion inspiration almost exclusively from the top
trending posts on Instagram, releasing 10 styles a
week that customers can order before they go into
production (Pallant & Sands, 2018). By creating
only products that customers have committed to
buying, Choosy avoids accumulating surplus stock
and leverages the upside of mass customization
while minimizing the downside.
Lily AI is another platform that assists in product
configuration in online settings. In particular, Lily
AI allows fashion retailers to encourage consumers
to ‘complete the look’ at checkout. Lily AI’s deep
understanding of shoppers’ choices about all
apparel categories enables it to create real-time
head-to-toe outfit suggestions that help retailers
to increase basket size at checkout. AI is also being
applied to product strategies within the store
environment. U.S. shopping service Instacart uses
ML to optimize in-store product selection,
equating to thousands of hours of labor savings
(Brandon, 2017). Fashion brand Levi’s is employing
algorithms to optimize how products are arranged
in store and to improve size availability. Likewise,
Nike is using geographical and behavioral data
from its app to inform store offerings, and it is
using clustering algorithms to provide advice on
which items should be displayed together
(McDowell, 2019). Product marketing is also being
fine-tuned with AI, with Samsung engaging Crimson
Hexagon’s AI-powered audience-insights platform
to understand what its existing and potential cus-
tomers are saying on social media (Sentence,
2018). In this way, the analysis of user-generated
conversations and associated images in social
channels assists in learning “how consumers
interact with their products, and thus how to
create marketing campaigns that they can relate
to” (Sentence, 2018).
3.6. Developing pricing strategy
Stage 6 involves developing a pricing strategy to
maximize sales. In developing a pricing strategy,
marketers decide how much to charge for products
and services, strive to understand consumer price
sensitivity, and map competitor pricing. AI can
assist in a number of ways, including estimating
consumer price elasticity, enabling dynamic pric-
ing (e.g., surge pricing), and detecting pricing
anomalies (including pricing errors, cases of fraud,
and nonprofitable customers). AI enables mar-
keters to track buying trends and determine more
competitive product price points in an attempt to
nudge customers at the point of decision
(Arevalillo, 2019). Amazon collects and analyzes
data at multiple points along the customer’s
journey, from prepurchase (including products
viewed or searched for, reviews, and page visits)
to purchase (including all purchase histories and
wish lists) and postpurchase (including returns and
service interactions). AI and ML assist Amazon in
collecting all of this customer data and under-
standing what shoppers are looking for and the
prices they are willing to pay (Ke, 2018).
Some firms are using dynamic pricing, supported
by big data and AI, to help gain a competitive
pricing advantage. In the hotel context, dynamic
pricing can allow the issue of underoccupancy to
be addressed by adjusting pricing to balance sup-
ply and demand and to maximize profit (O’Hear,
2017). To assist in pricing decisions, Airbnb em-
ploys AI and ML to help hosts make pricing de-
cisions about their property. Pricing is a complex
process for hosts given traditional demand factors,
such as seasonal changes, local events, and loca-
tion, as well as the fact that each listing exhibits
unique property characteristics (Hill, 2015). Airbnb
provides pricing assistance with an ML algorithm
that makes pricing suggestions for each date that a
host makes a property available. For brands that
want to ensure their pricing is competitive, AI can
develop a pricing index against competitors’ cat-
alogs and pricing, allowing relative price to be
benchmarked.
3.7. Developing channels and logistics
strategy
Stage 7 involves the development of a channel and
logistics strategy, including determining optimal
logistics, distribution, and product stocking. At
this stage, marketers strive to decide between
direct sales channels, whole channels, or retail
channels. In some instances, AI can provide access
to new channels to market. One social commerce
app, Browzzin, combines AI and visual-recognition
technology with the power of influencers to drive
sales as the app transforms posted images into
shoppable content (Dorfer, 2019). Shoppable-
content platforms like Browzzin and Pinterest’s
visual search feature are examples of deep
learning augmenting the shopping experience via
image classification. Such technologies allow con-
sumers to take pictures of things they see in
stores, on a commute, or at a friend’s, and then
How marketers can leverage AI 11
they make the items in the pictures shoppable. In
this way, consumers can shop anywhere or
anytime, which creates new opportunities for
companies to engage consumers outside of tradi-
tional channel locations. Similar developments will
also likely occur in the B2B sales process, which is
undergoing substantial transformations fueled by
advances in AI and ML (Paschen, Wilson, &
Ferreira, in press).
With regard to logistics strategy, marketing
managers are concerned with ensuring the delivery
of products so they are available to customers at
the right place and at the right time. From this
perspective, decisions need to be made to esti-
mate demand at particular locations, including at
the store level (considering historical and real-
time sales, real-time inventory, and in-store and
web-traffic data) and location level (considering
local competitors and demographics of local
catchment). AI allows marketers to predict and
optimize distribution, inventory, store displays,
and store layouts (both brick-and-mortar and on-
line). Today, retailers are able to employ AI-
informed planograms, or dynamic plans that
recommend the ideal number and placement of
inventory within a store (McDowell, 2019). Further
benefits to logistic management include cognitive
procurement and predictive merchandising, which
assist with stock and inventory management. At
Walmart, some stores have started testing auton-
omous robots that scan shelves for spaces that
need replenishing (McDowell, 2019). In the future,
it is feasible that robots and drones will transform
last-mile delivery.
3.8. Developing marketing communications
and influence strategy
Stage 8 involves the development of marketing
communications and influence strategy, with a
specific focus on serving customers the right pro-
motion at the right time. At this stage, marketers
work to create and enhance brand meaning in the
eyes of customers, as well as to inform them of
product offerings. This involves careful develop-
ment, targeting, and placement of communica-
tions in order to convey an effective message to
the correct set of customers while minimizing
costs. A diverse range of opportunities exists for AI
within the broad domain of marketing communi-
cations, including conducting AI-driven A/B ad
testing, contextual ad targeting, AI-optimized ad
retargeting, keyword bidding, and automation and
personalization of content creation. In terms of
data requirements, marketers should seek histor-
ical data to optimize placement and creation of
ads, as well as real-time data about customer
behavior at the point of purchase. Current appli-
cations at this stage include AI-powered interac-
tive ad placements (Sweeney, 2018) and the ability
to display ads in videos specific to the content
being watched (Caygill, 2017). For Black Friday
2018, LEGO engaged Watson Ads Omni to create AI-
powered interactive ads (Sweeney, 2018). The AI
system was trained with the knowledge of a wide
range of different LEGO products, with ads crafted
to consumers depending on their specific interests
and needs. The benefit of such an application is
that the brand can have meaningful, one-on-one
conversations with consumers along their paths
to purchase.
In terms of AI-generated content creation and
the development of personalized content, 20
th
Century Fox and the NBA provide interesting use
cases. In 2016, 20
th
Century Fox collaborated with
IBM Watson to create the first AI-created trailer for
the movie Morgan. Watson analyzed hundreds of
thriller and horror trailers to learn what aspects
make them suspenseful, and it suggested appro-
priately thrilling moments for inclusion in the
trailer (IBM, n.d.). Taking content creation one
step further, the NBA has worked with AI to
develop personalized content. In 2015, the NBA
partnered with U.S. sports-tech firm WSC Sports to
offer fans near-instant highlight clips from games
via NBA websites. Using AI, highlight packages
were developed for every player in a game,
enabling the NBA to deliver personalized highlights
to a global audiencedfor instance, sending clips of
Australian-born NBA stars to Australian viewers
who were previously underserved by traditional
broadcast footage (NBA, 2015).
3.9. Planning metrics and implementation
control
Finally, Stage 9 involves developing planning met-
rics and implementation control, and specifically
striving to identify and monitor performance
metrics and then taking any necessary corrective
actions. To this end, marketers use metrics to
assess how their efforts are working, to identify
problems, and to increase efficiency. Key aspects
of this task include identifying and measuring
relevant metrics and deciding how to respond to
abnormalities. In terms of data requirements, both
historical and real-time environmental sales and
marketing performance data can be used to di-
agnose and help predict corrective actions and
variability. Two key benefits of AI in planning and
implementation are that human operators are not
required to command or analyze outputs and that
12 C. Campbell et al.
AI works on a trial-and-error basis. Algorithms are
able to pick up detailed information by mimicking
the behavior of the human brain, and marketers
are able to understand, anticipate, analyze, and
act to solve problems.
A key way in which AI is being used at this stage
of the marketing function is in A/B testingdfor
instance, in the context of assessing advertising or
online features. A key benefit of A/B testing
underpinned by AL and ML is that websites, ads,
and other online assets can self-optimize in real
time as a result of AI assessment of behavior and
reactions to a multitude of different variations. ML
algorithms can continuously collect data and
deliver the optimized variations to individual users
in real time, allowing for the best performing
aspect to carry through. HSBC used this method to
drive 100% more click-throughs on their mobile
homepage (Adobe, 2019). The company tested AI-
driven dynamic content against static content on
its mobile app home page. The personalized re-
sults drastically outperformed their static peers in
terms of click-through rates to product pages. The
efficacy of a range of campaigns can be tested
with AI assistance. Many consumers use their
phones or other devices while they watch TV. If a
TV advertising campaign airs and customers use
devices in parallel to ad exposure, the immediate
response to the campaign can be assessed. For
example, with the data for when each individual
ad airs, marketing managers can assess metrics
relating to the specific call to action, such as site
traffic, and compare the airing times to user
response immediately.
4. Getting it right: From AI foundations
to an AI orientation
While AI holds many possibilities for marketers,
achieving its potential is not easy. A firm’s journey
from an AI foundation to an AI orientation is akin to
the DIKW Pyramid (Zins, 2007), which describes
the hierarchical relationship between Data, Infor-
mation, Knowledge, and Wisdom. At the base of
the pyramid, firms must first have or develop
certain foundational abilities to collect data. From
here, AI and ML functionality transforms useful and
relevant data into information. This information,
when blended with context, expertise, and intui-
tion, becomes knowledge. Finally, wisdom adds
value, which requires the mental function of
judgment (Wallace, 2007).
A data foundation requires firms to have robust
systems capable of tracking relevant data in real
time. They must also be able to store data for
historical analysis because the data requirements
for AI are quite high (Chui et al., 2018). While
collecting data may seem easy, it is important to
note that many IT systems are not designed for
data pulls. For instance, many advertising
agencies’ systems are designed to track an indi-
vidual client’s performance, not to extract data
from campaigns. Similarly, many businesses suffer
from data silos. Many systems do not use a com-
mon tracking ID number, or worse, they arrange
data in fundamentally different ways, necessi-
tating custom software for data sets to be merged.
While some forms of ML (e.g., reinforcement) can
operate without training data sets, accessing his-
torical data sets is valuable.
In developing an AI foundation, firms will also
need to be aware of the increasingly important
challenges of privacy and regulation. Consumers
are understandably growing more concerned about
what data is being collected from them and how it
is being used by marketers. In response, firms such
as Apple are proactively choosing to restrict what
consumer data is collected and how it is used.
While some workarounds are possible, such re-
strictions necessarily impose constraints on what
data scientists can potentially achieve with ML. A
similar concern lies in privacy-related regulation.
In Europe, the General Data Protection Regulation
(GDPR) came into effect in May of 2018. The
regulation requires the highest privacy settings by
default, unless a user explicitly consents to their
data being used. Such defaults restrict both the
amount and quality of data available for use in AI
applications. In the U.S., concern is also growing,
in part because of Facebook’s continued data
breaches and harvesting of data without consumer
consent (Doffman, 2019). U.S. regulation has yet
to follow the EU’s example, although it may if
consumer concern continues to grow. Further-
more, ML must also be monitored to ensure
compliance with antidiscrimination laws (Chui
et al., 2018; Newell & Marabelli, 2015). For
instance, in the U.S., lenders cannot use race-
based predictions of loan defaults as a reason for
loan denial, and in the EU, insurers are no longer
able to use statistical evidence about gender dif-
ferences to set premiums (Newell & Marabelli,
2015).
In addition to data, firms looking to leverage AI
also need to consider their human capital. While AI
requires skilled data scientists, other comple-
mentary functions are also necessary. While data
scientists are often highly skilled statisticians,
their expertise should be supplemented by deep
awareness of the firm, its markets, and its cus-
tomers in order to build optimal models. While ML
How marketers can leverage AI 13
algorithms are highly adaptable on their own,
correctly coding or transforming data can still lead
to significant improvements in performance. For
instance, while many data sets include a calendar
date, adding in a categorical variable denoting
which day of the week each day corresponds to
can be highly predictive of sales at some busi-
nesses, such as restaurants. As such, best practice
often marries data scientists with managers who
have deep business and strategic knowledge.
Software and software engineers are also impor-
tant considerations. While many open-source sta-
tistical computing programs have ML plug-ins
available, their use requires considerable exper-
tise to run, assess, and implement models
(Demandbase, 2016). Even more user-friendly op-
tions, such as the automated ML platform Data-
Robot, still require software engineers to integrate
resulting learnings into firm systems for on-the-fly
implementation. This reflects the broader choice
between buying or building IT architecture. It is
also important to be aware that ML models can
drift over time and require periodic recalibration
and retraining (Chui et al., 2018).
Once the foundational aspects of AI
implementationddata, human capital, and
softwaredare present, firms must still work to
develop an AI orientation within their broader cul-
ture. For some, such cultural shifts will require a
completely different approach to decision making,
from top down to bottom up and from long-range
planning to short-term reactions (Merendino et al.,
2018). This entails both philosophical shifts and
more pragmatic considerations. At a macro level,
business orientations need to shift toward a culture
of continual improvement through testing and
learning. While initial benefits from AI are often
modest, with continual improvement they can
compound over time into much larger gains. This
includes adopting a data-driven approach to mining
historical data for clues on paths forward, as well as
developing sets of testable possibilities that AI can
experiment upon. To evolve a business’s processes in
order to emphasize this constant mining of existing
data and ceaseless running of AI tests can sometimes
represent a large departure from existing organiza-
tional cultures. This requires training employees in
the basic capabilities and opportunities of AI so that
they can better identify possible business opportu-
nities, understand output, and act upon findings
(Chui et al., 2018; Demandbase, 2016). Internal
marketing programs to garner buy-in and share ideas
and successes can also be helpful.
While the success of developing an AI orienta-
tion often depends on a firm’s ability to foster
cross-functional collaboration, this is very difficult
to achieve. While considerable research explores
how to develop such competenciesdgenerally in
the context of developing a new product (Morgan
& Liker, 2006; Sethi, Smith, & Park, 2001)dthe
role of incentives is also an important consider-
ation. How AI resource costs and benefits are
apportioned can encourage different behaviors
among employees. For instance, if the budget for
AI expenditures is charged back to departments, it
might cause subtle resistances to usage, especially
in early stages when benefits are less clear. This
can be potentially mitigated if benefits from AI
cost savings accrue back to the individual de-
partments that implement them. Should such
savings be taken away, this might instead dissuade
use. Similarly, managers are encouraged to
consider how AI factors into the metrics by which
teams and departments are assessed. Rather than
simply using dollar amounts as targets, firms might
also set targets for efficiency gains. Likewise,
rather than evaluating immediate benefits, teams
and departments might be better evaluated in
terms of the opportunity or potential value their
changes can bring.
Because understanding and implementation of
AI is continuing to evolve, maintaining an AI
orientation requires constant monitoring of in-
dustry best practices. Managers may want to in-
crease funding for AI-related conferences and
educational opportunities, as well as encourage
sharing of insights across functional units. Careful
design of incentives and performance metrics can
help encourage participation in such efforts.
Finally, managers should lead by example, them-
selves staying abreast of best practices and
championing their own AI initiatives.
5. What’s next? Thoughts for the future
of AI
There is no doubt that AI is becoming increasingly
integrated into marketing practice, enabling
companies to reduce process times and engage
with individual consumers at scale. But in many
ways, AI is still in its infancy, and not all brands are
equally predisposed to implementing AI. Many
marketers likely fear relinquishing control to AI,
and across many industries, there are widespread
concerns that AI will take jobs away from workers.
But the rapid development of AI technologies will
see jobs change and adapt, though not necessarily
decline, to the evolving needs of companies. For
instance, as AI is responsible for more analytical
14 C. Campbell et al.
tasks, analytical skills will likely become less
important. In contrast, jobs requiring other skill
sets, such as intuition or empathy, are likely to
increase in importance (Huang & Rust, 2018). In
essence, to progress to the point where AI can add
wisdom, human judgment will be required.
With the multitude of opportunities for AI inte-
gration within marketing, some may fear that AI will
replace human work roles within the profession. But
at a broad level, the automation afforded by AI
technologies will allow marketers the ability to
invest their time in creativity rather than process.
For service agents, AI will result in an elevation in
tasks, allowing AI to improve the prioritization of
service agents’ work (Salesforce, 2019). At a more
granular level, AI affords marketers with new tools
to engage, satisfy, and retain customers.
The development of AI will also lead to a gap
between early adopters and laggards. Research
suggests that early adopters will likely be those
organizations with a strong digital base and, as
such, a higher propensity to invest in AI (Bughin
et al., 2017). For early adopters, there will be
additional challenges associated with forging a
new path in the largely uncharted application of AI
in business. These organizations may require un-
foreseen resources. For example, tech companies
like Facebook and Twitter have hired human-rights
directors to proactively address social and political
issues raised by emerging technology (Lomas,
2018). While AI laggards will not face such chal-
lenges, the customer experience they provide may
sufferdthough customers who want to resist AI or
who prioritize privacy may find laggard firms
attractive (Newell & Marabelli, 2015).
Regardless of the rate of AI adoption, any future
AI-marketing integration will be best served by
balancing AI and human intelligence. Given the
unique strengths and weaknesses of both AI and
humans, blending them together provides the
seamless end-to-end experience consumers expect
(Forrester, 2017). The most immediate integration
of AI and human intelligence will likely occur
within customer service via AI-assisted human
agents. By combining the speed of computer pro-
grams with the deep knowledge and understanding
of human service agents, companies can resolve
customer problems more quickly. Over time, it
may become increasingly hard to tell humans and
AI agents apart in service contexts.
Beyond operational marketing functions, AI will
likely effect drastic shifts in consumption practices
as we know them. Even today, societal changes
enabled by AI have shifted consumption behavior
in some categories. Now, rather than purchase
movies or music, consumers subscribe to Netflix or
Spotify, with these platforms offering a vast array
of titles and convenience, all curated to individual
users by intelligent recommendation systems. One
day, we may not own products in the traditional
sense but instead subscribe to them, and AI-
enabled platforms will deliver us food, clothing,
and other necessities as and when we need them
(Press, 2019). Such business models will continue
to advance the relevance of access over owner-
ship. These shifts will be important for businesses
to consider as they pursue the growing market of
consumers engaging in alternative forms of
consumption.
Acknowledgment
We gratefully acknowledge Dr. Taylor Larkin for
his machine-learning expertise.
References
Adobe. (2019 September 16). AI-powered personalization d
Above expectation, beyond imagination. Available at
https://www.adobe.com/insights/ai-powered-
personalization.html
Arevalillo, J. M. (2019). A machine learning approach to assess
price sensitivity with application to automobile loan seg-
mentation. Applied Soft Computing, 76, 390e399.
Baker, R. (2017 March 1). Domino’s to become an ‘artificial in-
telligence first’ business. Ad News. Available at http://
www.adnews.com.au/news/domino-s-to-become-an-
artificial-intelligence-first-business#0mPiDLEUG9Sg5FBA.99
Baumgartner, T., Hatami, H., & Valdivieso, M. (2016 May). The
new world of sales growth. McKinsey & Company. Available
at https://www.mckinsey.com/business-functions/
marketing-and-sales/our-insights/the-new-world-of-sales-
growth
Blueshift. (2018). Activating customer data for AI powered
marketing. Available at https://downloads.blueshift.com/
report-customer-data-for-ai/
Brandon, J. (2017 July 12). Instacart AI helps personal shoppers
buy groceries faster. VentureBeat. Available at https://
venturebeat.com/2017/07/12/instacart-ai-helps-personal-
shoppers-buy-groceries-faster/
Bughin, J., McCarthy, B., & Chui, M. (2017 August 28). A survey
of 3,000 executives reveals how businesses succeed with AI.
Harvard Business Review. Available at https://hbr.org/
2017/08/a-survey-of-3000-executives-reveals-how-busi-
nesses-succeed-with-ai
Caygill, D. (2017 November 20). Six trends brands should know
about for 2018 and the tech they need to craft responses.
Campaign. Available at https://www.campaignlive.co.uk/
article/six-trends-brands-know-2018-tech-need-craft-
responses/1450488
Chaffey, D. (2004). E-business and e-commerce management.
London, UK: Financial Times-Prentice Hall.
Chaffey, D., & Smith, P. R. (2012). Emarketing excellence:
Planning and optimizing your digital marketing (4
th
ed.).
Abingdon, UK: Routledge.
How marketers can leverage AI 15
Choudhury, S. R. (2016 May 24). Softbank’s Pepper robot gets a
job waiting tables at Pizza Hut. CNBC. Available at https://
www.cnbc.com/2016/05/24/mastercard-teamed-up-with-
pizza-hut-restaurants-asia-to-bring-robots-into-the-pizza-
industry.html
Chui, M., Manyika, J., Miremadi, M., Henke, N., Chung, R.,
Nel, P., et al. (2018, April). Notes from the AI frontier: In-
sights from hundreds of use cases. New York, NY: McKinsey
Global Institute.
Columbus, L. (2018 May 14). 77% of marketing execs see AI
adoption growing this year. Forbes. Available at https://
www.forbes.com/sites/louiscolumbus/2018/05/14/77-of-
marketing-execs-see-ai-adoption-growing-this-year/
#51c6e4267ef8
Cui, G.,Wong, M. L., & Lui,H. K. (2006). Machinelearning for direct
marketing response models: Bayesian networks with evolu-
tionary programming. Management Science, 52(4), 597e612.
Das, S., Doppa, J. R., Kim, D. H., Pande, P. P., & Chakrabarty, K.
(2015). Optimizing 3D NoC design for energy efficiency: A
machine learning approach. In Proceedings of the 2015
IEEE/ACM international conference on computer-aided
design (pp. 705e712). Austin, TX: IEEE.
Davenport, T. H., & Kirby, J. (2016). Just how smart are
smart machines? MIT Sloan Management Review, 57(3),
21e25.
Demandbase. (2016 December 13). 80 percent of all marketing
executives predict artificial intelligence will revolutionize
marketing by 2020. Available at https://www.demandbase.
com/press-release/marketing-executives-predict-artificial-
intelligence-will-revolutionize-marketing-2020/
Doffman, Z. (2019 April 18). 1.5m users hit by new Facebook
privacy breach as extent of data misuse exposed. Forbes.
Available at https://www.forbes.com/sites/zakdoffman/
2019/04/18/facebook-illegally-harvested-data-from-1-5m-
users-as-it-leveraged-its-data-machine/#31649c436a2e
Dorfer, S. (2019 May 31). Shoppable content: AI app Browzzin.
Available at https://www.stylus.com/kppfqb
Duhigg, C. (2012). How companies learn your secrets. New York
Times Magazine. Available at https://www.nytimes.com/
2012/02/19/magazine/shopping-habits.html?
pagewantedZ1&_rZ1&hp
Dunwoodie, B. (2018 July 12). How AI is impacting the voice of
the customer landscape. CMS Wire. Available at https://
www.cmswire.com/customer-experience/how-ai-is-
impacting-the-voice-of-the-customer-landscape
Elliott, C. (2018 August 27). Chatbots are killing customer ser-
vice. Here’s why. Forbes. Available at https://www.forbes.
com/sites/christopherelliott/2018/08/27/chatbots-are-
killing-customer-service-heres-why/#466539aa13c5
Evans, G. L. (2017). Disruptive technology and the board: The
tip of the iceberg. Economic and Business Review, 3(1),
205e223.
Fluss, D. (2017). The AI revolution in customer service. Desti-
nationCRM. Available at https://www.destinationcrm.com/
Articles/ReadArticle.aspx?ArticleIDZ115528
Forrester. (2017). Forrester study reveals benefits of artificial
intelligence with the human touch. Genesys. Available at
https://www.genesys.com/collateral/forrester-study-
reveals-benefits-of-artificial-intelligence-with-the-human-
touch
Hill, D. (2015). How much is your spare room worth? IEEE
Spectrum, 52(9), 32e58.
Huang, M. H., & Rust, R. T. (2018). Artificial intelligence in
service. Journal of Service Research, 21(2), 155e172.
Hurwitz, J. S., Kaufman, M., & Bowles, A. (2015). Cognitive
computing and big data analytics. New York, NY: John Wiley
& Sons.
IBM. (n.d.). The quest for AI creativity. Available at https://
www.ibm.com/watson/advantage-reports/future-of-
artificial-intelligence/ai-creativity.html
Javelosa, J. (2017, March 30). Major firm announces it’s
replacing its employees with AI. Futurism. Available at
https://futurism.com/major-firm-announces-its-replacing-
its-employees-with-a-i
Jordan, M. I., & Mitchell, T. M. (2015). Machine learning:
Trends, perspectives, and prospects. Science, 349(6245),
255e260.
Ke, W. (2018, November 2). Power pricing in the age of AI and
analytics. Forbes. Available at https://www.forbes.com/
sites/forbesfinancecouncil/2018/11/02/power-pricing-in-
the-age-of-ai-and-analytics/#6ce8b9fc784a
LeCun, Y., Bengio, Y., & Hinton, G. (2015). Deep learning. Na-
ture, 521(7553), 436e444.
Lomas, N. (2018, September 16). Facebook is hiring a director of
human rights policy to work on “conflict prevention” and
“peace-building.” TechCrunch. Available at https://
techcrunch.com/2018/09/16/facebook-is-hiring-a-director-
of-human-rights-policy-to-work-on-conflict-prevention-and-
peace-building/
Marr, B. (2018 February 14). The key definitions of artificial
intelligence (AI) that explain its importance. Forbes. Avail-
able at https://www.forbes.com/sites/bernardmarr/2018/
02/14/the-key-definitions-of-artificial-intelligence-ai-that-
explain-its-importance/#4eebeecc4f5d
Martı
´
nez-Lo
´pez, F. J., & Casillas, J. (2013). Artificial
intelligence-based systems applied in industrial marketing:
An historical overview, current and future insights. Indus-
trial Marketing Management, 42(4), 489e495.
McCorduck, P. (2009). Machines who think: A personal inquiry
into the history and prospects of artificial intelligence.
Natick, MA: AK Peters.
McDowell, M. (2019 March 12). Stores get smart about AI. Vogue
Business. Available at https://www.voguebusiness.com/
technology/artificial-intelligence-physical-stores-kering-
nike-alibaba
Merendino, A., Dibb, S., Meadows, M., Quinn, L., Wilson, D.,
Simkin, L., et al. (2018). Big data, big decisions: The impact
of big data on board level decision-making. Journal of
Business Research, 93,67e78.
Moore, M. (2018 July 14). Wimbledon 2018: How IBM Watson
is serving up the best viewer experience. TechRadar. Avail-
able at https://www.techradar.com/news/wimbledon-
2018-how-ibm-watson-is-serving-up-the-best-viewer-
experience
Morgan, J. M., & Liker, J. K. (2006). The Toyota product
development system. New York, NY: Productivity Press.
Narrative Science. (2018 January 17). Artificial intelligence (AI)
adoption grew over 60% in the last year: Executive survey
shows almost two-thirds of enterprises harnessed AI in
2017. Available at https://narrativescience.com/company-
annoucements/artificial-intelligence-ai-adoption-grew-
over-60-in-the-last-year/
NBA. (2015). NBA teams up with WSC Sports Technologies to pro-
vide next-gen video highlights to fans worldwide. Available at
http://pr.nba.com/nba-wsc-sports-technologies-partnership
Negnevitsky, M. (2004). Artificial intelligence: A guide to
intelligent systems (2
nd
ed.). New York, NY: Addison-Wesley.
Newell, S., & Marabelli, M. (2015). Strategic opportunities (and
challenges) of algorithmic decision-making: A call for action
on the long-term societal effects of ‘datification’. The
Journal of Strategic Information Systems, 24(1), 3e14.
Oculus360. (2018 March 1). Using AI to analyze your brand’s
category. Available at https://www.o360.ai/blog/using-ai-
to-analyze-a-brand-category?hsCtaTrackingZ1a0f5798-
16 C. Campbell et al.
592d-445c-b8b3-31e122a70676%7Cf1c6b498-0a32-4ced-
9a9c-dd8f3ceb236b
O’Hear, S. (2017 December 13). The price is right! Pace raises
£2.5M to automate hotel room pricing based on demand.
TechCrunch. Available at https://techcrunch.com/2017/12/
13/pace-pricing/
Pallant, J., & Sands, S. (2018 September 12). The paradox of
choice. Why made-to-order might not solve the fashion
industry’s problems. The Conversation. Available at https://
theconversation.com/the-paradox-of-choice-why-made-to-
order-might-not-solve-the-fashion-industrys-problems-
102442
Paschen, U., Pitt, C., & Kietzmann, J. (2020). Artificial intelli-
gence: Building blocks and an innovation typology. Business
Horizons, 63(2) (XXXeXXX).
Paschen, J., Wilson, M., & Ferreira, J. (2019). Collaborative
intelligence: How human and artificial intelligence create
value along the B2B funnel. Business Horizons, 63(3) (in
press).
Ponce de Leon, S. (2019 January 12). At CES, AI robots can read
your emotions. Grit Daily. Available at https://gritdaily.
com/ai-emotion-bots/
Power, B. (2017 May 30). How Harley-Davidson used artificial
intelligence to increase New York sales leads by 2,930%.
Harvard Business Review. Available at https://hbr.org/
2017/05/how-harley-davidson-used-predictive-analytics-to-
increase-new-york-sales-leads-by-2930
Press, G. (2019 December 9). 120 AI predictions for 2019. For-
bes. Available at https://www.forbes.com/sites/gilpress/
2018/12/09/120-ai-predictions-for-2019/#43092141688c
Reavie, V. (2018, August 1). Do you know the difference be-
tween data analytics and AI machine learning? Forbes.
Available at https://www.forbes.com/sites/
forbesagencycouncil/2018/08/01/do-you-know-the-
difference-between-data-analytics-and-ai-machine-
learning/#5c50edac5878
Rust, R. T., & Huang, M. H. (2014). The service revolution and
the transformation of marketing science. Marketing Science,
33(2), 206e221.
Salesforce. (2017). State of marketing: Insights and trends from
3,500 global marketing leaders. Available at https://www.
salesforce.com/content/dam/web/en_us/www/assets/
pdf/datasheets/salesforce-research-fourth-annual-state-of-
marketing.pdf
Salesforce. (2019 April 9). State of service: 2019’s changing
customer service trends. Available at https://www.
salesforce.com/au/blog/2019/04/state-of-service–2019-s-
changing-customer-service-trends.html
Sentence, R. (2018 June 28). How Samsung uses social listening
for product marketing and sentiment analysis. Available at
https://econsultancy.com/how-samsung-uses-social-
listening-for-product-marketing-sentiment-analysis/
Sethi, R., Smith, D. C., & Park, C. W. (2001). Cross-functional
product development teams, creativity, and the innova-
tiveness of new consumer products. Journal of Marketing
Research, 38(1), 73e85.
Sweeney, E. (2018 October 2). IBM’s interactive AI ads reach
more sites, brands. Industry Dive. Available at https://
www.marketingdive.com/news/ibms-interactive-ai-ads-
reach-more-sites-brands/538558
Teboul, W. (2018 July 20). Why use machine learning instead of
traditional statistics? Available at https://
towardsdatascience.com/why-use-machine-learning-
instead-of-traditional-statistics-334c2213700a
Vilar, H. (2019 March 12). UK’s Nationwide analyses customer
interactions with SAS. Fintech Futures. Available at https://
www.bankingtech.com/2019/03/uks-nationwide-analyses-
customer-interactions-with-sas/
Wallace, D. P. (2007). Knowledge management: Historical and
cross-disciplinary themes. Santa Barbara, CA: Libraries
Unlimited.
Wolska, K. (2017 May 19). From mobile first to AI first dGoogle
I/O 2017 conference. Available at https://medium.com/
appchance/from-mobile-first-to-ai-first-google-i-o-2017-
conference-c93247d8c234
Wood, M. B. (2011). The marketing plan handbook. New York,
NY: Pearson.
Zins, C. (2007). Conceptual approaches for defining data, infor-
mation, and knowledge. Journal of the American Society for
Information Science and Technology, 58(4), 479e493.
How marketers can leverage AI 17
